[MUSIC] So the way we're gonna learn about this
data to intelligence pipeline is by examining a number of case studies that
are gonna ground the methods that we present in real world applications. And that's one of the really
unique features of this course. In our first case study, we're gonna
look at predicting house values. So, the intelligence we're deriving is a value associated with some
house that's not on the market. So, we don't know what it's value is and
we wanna learn that from data. And what's our data? Well in this case, we're gonna look at
other houses and look at their house sales prices to inform the house value
of this house we're interested in. And in addition to the sales prices, we're gonna look at other
features of the houses. Like how many bedrooms the houses have. Bathrooms, number of square feet,
and so on. And what we're gonna do,
our machine learning method is something that's gonna relate the house
attributes to the sales price. Because if we can learn this model, this
relationship from our house level features to the observed sales price, then we can
use that for predicting on this new house. We take its house attribute and
predict its house sales price. And this method is called regression. In our second case study,
we're gonna explore a sentiment analysis task where we have reviews
of some restaurants. So for example in this case, it says the
sushi was awesome, the food was awesome, but the service was awful. And we wanna take this review and be able to classify whether
it had positive sentiment. It was a good review, thumbs up or
a negative sentiment, thumbs down. And so how are we gonna do this? Well, we're gonna look at
a lot of other reviews. So, we're gonna look at the text of
the review and the rating of the review. In order to understand what's
the relationship here, for classification of this sentiment. So, for example in this case, maybe we might analyze the text of this
review in terms of how many times it uses the word awesome versus how many
times it uses the word awful. And from these other reviews that we have,
we're gonna learn some decision boundary between based on the balance of usage of
these words whether it's a positive or negative review. And the way we learn that from these other
reviews is based on the ratings associated with that text. And so this method is called
a classification method. In our third case study, we're gonna do
a document retrieval task where here, what we wanna do, the intelligence
we're deriving is an article or a book or something like this
that's of interest to our reader. And the data that we have is a huge
collection of possible articles that we could recommend. And what we're gonna do,
in this case, is we're gonna try and find structure in this data based
on groups of related articles. Such as, maybe there's a collection of
articles about sports and world news and entertainment and science. And if we find this structure and
annotate our corpus, our collection of documents
with these types of labels which we don't have ahead of time,
we're trying to infer this from the data. Then we can use this for very rapid
document retrieval because if I'm sitting here currently reading some article
about world news, then maybe, if I wanna retrieve another article, I already
know which articles to search over. And this type of approach
is called clustering. In our fourth case study, we're gonna do this really interesting
thing that's called collaborative filtering that's had a lot of impact
in many domains in the last decade. Specifically, we're gonna look
at doing product recommendation, where you take your past purchases and
trying to use those to recommend some set of other products
you might be interested in purchasing. So in this case, the data that we're
gonna use to derive this intelligence for product recommendation is we'd like to
understand what's the relationship between what you bought before and
what you're likely to buy in the future. And to do this, we're gonna use
other users' purchase histories. And possibly, features of those users. But the key idea here is we're
gonna take this data and we're gonna arrange it into this
customers by products matrix where the squares here indicate products
that a customer actually purchased. So those are products that
are liked by that customer. And from this matrix,
we're gonna learn features about users and features about products. And once we learn those
features about users and products from this data
that I've described. We can think about using those features
to see how much agreement there is between what the user likes,
different attributes the user likes and whether the product is actually
about those attributes. So in the example I'm showing here,
maybe a user is a mom, has certain features that are similar
to other users that are also moms. And from that,
we can infer things about products. What are attributes about? For example, baby products
that are of interest to moms. And we're using that information
to form our recommendations. And this type of approach
going from this matrix, this customers products matrix into
these learned features about users and products is called matrix factorization. Okay, well in our final case study, we're gonna look at a visual
product recommender. So here, our data is somebody's gonna
go to the web and they're gonna input, not text, but an image. They're gonna put an image like of
a black shoe, or a black boot, or a high heel, or some docker shoe,
or running shoe. And what they want is they want
a set of results of shoes that might also be of interest to them. So, shoes that are visually similar
to the picture that they have. And they wanna be able to search
over those to purchase this item. And the way we're gonna do this, to be able to go from an image to
a set of related images is we need to have very good features about that image
to find other images that are similar. And the way we're gonna derive those
really detailed features is something called deep learning. So in particular, we're gonna look at
these neural networks where every layer of the neural network provides more and
more descriptive features. So in the little example we show here, the first layer might just detect in
the image things like different edges. Whereas with one we get to the second
layer, we start detecting corners and more interesting features like that. And as you go deeper and
deeper in these layers, you get more intricate features arising. So as you see, we're gonna walk through
a series of real-world case studies, real-world problems and real-world
solutions using machine learning. And through this, we're gonna explore a series of methods
that have a lot of power out there. And are gonna allow you to be able to
develop and deploy new machine learning techniques on new problems that aren't
the exact case studies we used. But the case studies will allow us
to really ground the methods that we're describing with things
that are very interpretable. [MUSIC]